- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi all,
I wanted to compare the LU factorization time done by DSS and PARDISO on a quad core Intel CPU in MS VS 2008. In PARDISO I have set the iparm(2)=3 to use parallel threads, however I am not seeing any acceleration in compare with DSS or even when iparm(2)=0 or 2.
If PARDIOS is faster than DSS? and if DSS is using parallel threads by default?
Thanks.
Link Copied
4 Replies
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Vahid,
you won't see any speed up bacause DSS is just a simplier interface to Pardiso. Core is the same.
A.
you won't see any speed up bacause DSS is just a simplier interface to Pardiso. Core is the same.
A.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
exactly,DSS is just different, simplified interface which provides the same efficiency of calculations.
--Gennady
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
You mean both DSS and PARDISO are using all 4 cores of the quad core CPU by default? We don't need to set anything for this purpose, right?
Thank you.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
It depends of version of MKL you are using. For the latest (10.2 and 10.3) versions that's correct.
--Gennady
Reply
Topic Options
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page